Browser-based analytics for GitHub Copilot Enterprise usage and AI-credit cost data.
Live: kinncj.github.io/GitHub-Copilot-Enterprise-Dashboard
The dashboard reads GitHub's two Copilot exports. Drop either (or both); the file type is detected on upload, and when both are loaded a tab switches between them. They are analysed separately.
Team usage and productivity from the Copilot Enterprise API export.
- 14 charts — activity timelines, acceptance rate trends, user rankings, IDE/language/feature/model breakdowns
- KPI cards — total users, generations, acceptance rate, lines of code, estimated value
- Feature adoption — which Copilot features (Code Completion, Edit Mode, Chat · Agent, etc.) the team uses
- Insights — power users, high-efficiency users, spotlight users (top code producers), weekly trend, quota alerts
- Data table — per-user aggregated view, sortable and filterable
- Exports — CSV and PNG per chart, consolidated NDJSON
AI-credit consumption and cost from the billing export.
- Cost & credits — credits over time, top users, by model (auto-routed vs explicit), by org/cost-center/SKU
- Budget & burn rate — month-end projection at the user, org, and enterprise level, against each seat's monthly quota
- License config — enter real licenses per tier so org/enterprise budgets count idle seats the export leaves out
- Pooled overage — included credits pool across the billing entity, so projected overage is total consumption beyond the total allowance; heavy users are absorbed by the shared pool
- Credits priced at $0.01 each (GitHub's rate); projections under 7 days of data are marked preliminary
Nothing leaves the browser. No backend, no API calls, no tracking. Safe for enterprise usage and billing data.
npm install
npm run dev # http://localhost:3000Drop a Copilot Enterprise NDJSON export or an AI Usage Report CSV onto the upload zone — the type is detected automatically. Multiple NDJSON files merge into one date range (handy for overlapping 28-day rolling exports); load a CSV alongside them to get the cost and budget views.
| Layer | Tool |
|---|---|
| UI | React 18 + JSX |
| Build | Vite 5 |
| Charts | Chart.js 4 |
| Styles | Tailwind CSS 3 |
| Unit tests | Vitest |
| E2E tests | Playwright |
| Deploy | GitHub Pages via GitHub Actions |
Clean Architecture with SOLID principles. Three layers:
app/domain/— pure JS, no DOM. Two parallel pipelines: the activity one (data/,filtering/,insights/,export/) and the AI-usage one (aiusage/— parser, aggregator, filtering, insights, budget, export), plusdata/detect.jsto route uploads.app/state/useAppState.js— React hook that orchestrates both pipelines and holds the active view.app/presentation/— React components and Chart.js renderers (components/aiusage/andcharts/aiusage/for the cost dashboard).
Domain modules have no imports from the presentation or state layer — fully testable in Node without a browser.
See docs/architecture.md for diagrams.
| Doc | What's in it |
|---|---|
docs/data.md |
Every data point, all 14 charts, Data Explorer columns, insights, value calculation, the AI Usage report + budget/burn-rate model |
docs/architecture.md |
Layer diagrams, data flow, NDJSON schema, SOLID table |
docs/development.md |
How to add charts, insights, KPIs, filters, exports |
docs/configuration.md |
CONFIG thresholds, FEATURE_LABELS, value calculation |
docs/testing.md |
Unit + E2E test layout, how to write tests |
docs/deployment.md |
GitHub Pages pipeline, base path setup |
AGENTS.md |
Agent/Claude Code guidance, SOLID constraints, entry points |
GNU Affero General Public License v3.0 — see LICENSE.
Copyright © 2026 Kinn Coelho Juliao